relax SVN archive

Fix for the value.read() user function.

The directory and 'sep' args were being ignored.

Changed a RelaxError so it is more appropriated.

This is as proposed in the commit for r8137.

Modified the autodetection code for the generic format.

This now recognizes the most generic format as in
'test_suite/shared_data/peak_lists/generic_intensity2.txt'.

Modified the check for the spectrum_id being a string, a list is now accepted as well.

This is needed to implement the list of spectrum_id as in 'test_suite/system_tests/peak_lists.py'
(test_read_peak_list_generic2() function).

Should the RelaxStrError be changed to something like a RelaxStrListError ?

Complete fix for bug #12679 (https://gna.org/bugs/index.php?12679).

This bug was reported by Michael Funk (mifu att nmr dot mpibpc dot mpg dot de).

The last issue was that when there multiple graphs, one per spin, the spin labels for the graphs
were set to the x-axis values rather than the spin_id.

Partial fix for bug #12679 (https://gna.org/bugs/index.php?12679).

This bug was reported by Michael Funk (mifu att nmr dot mpibpc dot mpg dot de).

The problem was that the data for the graph was numpy arrays, and only lists were being checked for.

Bug fix for the temp dir removal in the relaxation curve-fitting system tests.

The loading of program state was resetting the relax data store, so now the name is also kept in
self.

Added a missing import causing a number of downstream system tests to fail.

Added code to check the Grace files created by the 1UBQ_relax_fit.py system test script.

Expanded the system test for both bug #12670 and bug #12679.